500 
ROBINSON. 
staminodes are present. Field as well as herbarium experience has 
testified to the accuracy of this statement regarding the involucre, though 
the specimens so seen were immature. If these are retained in Elatos- 
tema, no one character is left upon which to maintain, the separation of 
the genera. Furthermore, if An&rosyce is kept as a subgenus instead of 
a genus, the best of all the characters is lost. 
It is significant that in Weddell’s enumeration of the species of 
Elatostema in the Prodromus, writing there with his fullest knowledge 
of the plants, the only species definitely recorded as having a 5-parted 
pistillate perianth is E. rostratum Miq., to which he there reduced E. ma- 
nillense as a variety. He states that he had seen only staminate flowers of 
the latter : how very accurately he interpreted its alliance is shown by the 
fact that our sheet of Cuming 786 , on which it was based, has 1 pistillate 
flowers, and they have the inflorescence and perianth of E. rostratum. 
In other words, they have a very distinctive habit. There is little room 
for doubt that the very group under consideration is responsible for the 
exception made by Weddell in his statement of the number of perianth- 
divisions in Elatostema. 
The group, which occasioned Hallier’s difficulties, is similar but 
somewhat different, for no one of the 6 Bornean species described by 
him under his subgenus Pellionia is typical of the genus of that name. 
The pistillate flowers are described in 3 species as in capitula, or “subcapi- 
tatim congesti,” the perianth is 5-parted: the other species are clearly 
in the same alliance. But these differ from the E. manillense group in 
that the staminate inflorescence is long-pedunculate, though the flowers 
are not collected in an involucrate receptacle. We have a similar ease in 
Merrill 5269, from Bucas, an island northeast of Mindanao. The dif- 
ference seems no more Serious than that between the sessile and peduncled 
receptacles of Elatostema, or between the lax and the condensed cymes 
of different species of Pellionia. In these aberrant cases, the pistillate 
inflorescence is so condensed that it has the appearance of a receptacle, 
but the most ordinary care is sufficient to show that none is present. 
Neither of these two groups can be placed in Elatostema , as they lack 
its two distinctive characters, though they most nearly resemble it in 
habit, while sufficiently distinct even in that. If existing definitions are 
to be followed, they more nearly go into Pellionia, and it may well be 
a matter, of opinion whether the conception df that genus should be 
enlarged to include them, in which case they would make a distinct' sub- 
genus 'with two definite divisions, or whether it is better to place them 
by themselves. The latter course has here been followed, owing to the 
distinctive perianth, the greater condensation of at least the pistillate 
inflorescence, and the different habit. 
There is another consideration of a somewhat different nature. There 
is much temptation to consider Procris the most primitive genus of the 
